Mike Chambers wrote: > On Mon, 2007-05-28 at 09:58 +0800, Ed Greshko wrote: > >> IMHO, it is not a good idea to upgrade a system from a released/stable >> version of Fedora to a Beta/RC version of the next release. (Actually that >> goes for almost any OS.) That path is not formally tested and you may find >> yourself in dependency hell or in a very confused state. > > You need to rethink what you just said. It's *not* a good idea to > upgrade from a released version to a beta/rc? That path *is* formally > tested and is *a wanted test* to do during the development of the next > release. How else are upgrades tested? I am going to assume you mean a > live production system and not a testing/beta machine that is doing it? > Because that is part of the test, to upgrade from an earlier version to > the latest, which means doing it against a beta or rc. I don't think I need to rethink it. It is much more time saving to test each beta individually and then when you have your final RC test an upgrade from a formal previous release to a formal final release or final RC. If any problems are found at that point you fix those and then release. It is inefficient to resolve upgrade issues in multiple beta stages that may or may not exist when you get further down the line. >> For example, the RC2 may have version 6 of a particular library or >> application. Then, in testing, it is decided that 5 was more stable than 6 >> and when the released version comes out it has 5 but you already have 6 and >> the upgrade patch is not designed to go backwards. > > Above would be correct, but again depending on what exactly you are > referring to by upgrading, or at least which system (production? > beta/test machine?). I am referring to upgrading from a formal/supported release to a new formal/supported release.
